An example is intelligence, which constitutes the construct and can be observed through the measurement of variables observed as verbal and quantitative reasoning test scores, for instance, among other measurable indicators. In these models, the types of variables are distinguished according to their measurement or role in the model: (i) latent variables: also known as constructs, factors, concepts, or conceptual variables-they are the model features of direct interest, but they are unobservable elements that can only be inferred from those observed (ii) observed variables, also called indicators, inputs, or simply measures, and are distinguished because they can be measured and are known or thought to be related to the latent concepts. This method originated in the context of genetics, to examine the joint effect of one or more independent variables, which were represented in a path diagram, which is why it is also sometimes called broadly, path analysis. Structural equation modeling (SEM) is a multivariate statistical technique that allows researchers to estimate and test causal relationships. This example is useful if we want to understand what drives the intention to shop online.ġ Partial Least Square Structural Equation Modeling (PLS-SEM) The last variable, EE, is the degree of ease associated with consumers’ use of Internet for shopping. FC refer to consumers’ perceptions of the resources and support available to shop online. SI is the extent to which consumers perceive that importance others (e.g., family and friends) believe they should use Internet for shopping. PE is defined as the degree to which e-commerce will provide benefits to consumers for shopping. According to the UTAUT Model, BI is an indicator of how people are willing to shop online. Specifically, we test whether Behavioral Intention (BI) is predicted by Performance Expectancy (PE), Effort Expectancy (EE), Social Influence (SI), and Facilitating Conditions (FC).
We illustrate PLS-SEM using a subset of 2017 Customer Behavior in Electronic Commerce Study in Ecuador. This example shows in which situations researchers should use this technique with respect to other predictive multivariate techniques. Partial Least Squares Structural Equation Modeling (PLS-SEM) is useful when the research needs to predict a set of dependent variables from a large set of independent variables (Abdi, 2007).